Hosted MCP with OAuth
Use `https://mcp.apify.com` and authenticate in the browser. This is the recommended setup when the client supports remote MCP.
←Back to Skill DirectorySuggest update
Actor MCP
Apify MCP Server
Apify MCP Server turns thousands of Apify Actors into a single MCP-accessible tool layer. In Claude Code, that means one connection can expose actor search, actor execution, documentation lookup, and structured run outputs without building custom glue for every scraping or browser automation task.
The most important Apify design choice is hosted versus local setup. Apify recommends the hosted `mcp.apify.com` path with OAuth because it is fast and secure, but also documents a local stdio path for clients that are not ready for remote MCP servers.
Local stdio fallback
Run `npx -y @apify/actors-mcp-server` with `APIFY_TOKEN` when your MCP client prefers local stdio or cannot use remote HTTP.
Tool selection
Apify lets you load just the tools or Actors you need, which is critical for keeping tool choice clean in Claude Code.
Actor-driven workflows
Use the server to search Actors, inspect schemas, call Actors, and retrieve structured results for scraping and automation tasks.
Setup Paths
How to Set Up Apify MCP Server
Hosted server (recommended)
{
"mcpServers": {
"apify": {
"url": "https://mcp.apify.com"
}
}
}You can also pass `Authorization: Bearer <APIFY_TOKEN>` in headers if you do not want to use the OAuth sign-in flow.
Local stdio fallback
{
"mcpServers": {
"actors-mcp-server": {
"command": "npx",
"args": ["-y", "@apify/actors-mcp-server"],
"env": {
"APIFY_TOKEN": "YOUR_APIFY_TOKEN"
}
}
}
}Apify documents Node.js v18+ as the baseline requirement for local setup and recommends always using the latest package version.
Example tool selection
https://mcp.apify.com?tools=actors,docs,apify/rag-web-browser https://mcp.apify.com?tools=apify/instagram-scraper,apify/google-search-scraper npx @apify/actors-mcp-server --tools actors,docs,apify/web-scraper
Hosted first
Default to the hosted server when your client supports Streamable HTTP. It is the shortest path to a production-usable connection.
Limit tools early
Trim the tool list to the actors and docs surface you actually need so Claude does not see a noisy actor catalog on every turn.
Verify one actor end to end
Search one actor, inspect its schema, run it once, and verify output before expanding the server into a bigger workflow.
Execution Brief
Use this page as a rollout checklist, not just reference text.
Tool Mapping Lens
Organize Tools by Workflow Phase
Catalog-oriented pages work best when users can map discovery, evaluation, and rollout in a clear path instead of reading an undifferentiated list.
- Define the job-to-be-done first
- Group tools by stage
- Prioritize by adoption friction
Actionable Utility Module
Skill Implementation Board
Use this board for Apify MCP Server before rollout. Capture inputs, apply one decision rule, execute the checklist, and log outcome.
Input: Objective
Deliver one measurable improvement with apify mcp server
Input: Baseline Window
20-30 minutes
Input: Fallback Window
8-12 minutes
| Decision Trigger | Action | Expected Output |
|---|---|---|
| Input: one workflow objective and release owner are defined | Run preview execution with fixed acceptance criteria. | Go or hold decision backed by repeatable evidence. |
| Input: output quality below baseline or retries increase | Limit scope, isolate root issue, and rerun controlled test. | One confirmed correction path before wider rollout. |
| Input: checks pass for two consecutive replay windows | Promote to broader traffic with fallback path active. | Stable rollout with low operational surprise. |
Execution Steps
- Record objective, owner, and stop condition.
- Execute one controlled preview run.
- Measure quality, latency, and correction burden.
- Promote only when pass criteria are stable.
Output Template
tool=apify mcp server objective= preview_result=pass|fail primary_metric= next_step=rollout|patch|hold
What Is Apify MCP Server?
Apify MCP Server is Apify's official MCP integration for exposing Apify platform capabilities to AI clients. Rather than adding one scraper at a time as separate tools, you connect one server and gain access to Actor discovery, Actor execution, run results, and documentation retrieval from the Apify ecosystem. That makes it one of the most flexible MCP servers for scraping and automation-heavy teams.
The hosted server at `https://mcp.apify.com` is the primary path Apify recommends. It supports Streamable HTTP with OAuth and adds output schema inference for structured Actor results. That schema support matters because agents can understand expected result fields before calling an Actor, which improves reliability and lowers trial-and-error during execution.
Local stdio still matters because not every client handles remote MCP equally well. Apify documents a clean fallback using `@apify/actors-mcp-server` and `APIFY_TOKEN`. For Claude Code users, the right choice depends on whether you value fast hosted setup or the predictability of a local process tied to your workstation.
How to Calculate Better Results with apify mcp server
Start with the hosted path unless you have a concrete reason not to. Apify explicitly recommends `https://mcp.apify.com` with OAuth because it avoids manually handling API tokens inside client configs and keeps the setup fast. If your client cannot use remote HTTP smoothly, move to the local stdio server with `npx -y @apify/actors-mcp-server` and `APIFY_TOKEN` in the environment.
Next, trim the tool surface. Apify documents query-parameter based tool selection for hosted use and `--tools` for local use. This is not an optional optimization. If your workflow only needs actor search, docs, and one or two named Actors, load only those. Leaving the entire default surface available increases noise and weakens tool selection quality inside Claude Code.
Finally, test one real Actor workflow before declaring setup complete. A strong smoke test is searching Actors, fetching actor details to inspect schema and pricing, then calling a single Actor and verifying the output. If that full path works, the MCP integration is genuinely usable rather than just technically connected.
Treat this page as a decision map. Build a shortlist fast, then run a focused second pass for security, ownership, and operational fit.
When a team keeps one shared selection rubric, tool adoption speeds up because evaluators stop debating criteria every time a new option appears.
Worked Examples
Example 1: Web extraction inside a coding workflow
- Developer connects Claude Code to Apify MCP using the hosted server.
- Claude searches the Actor Store, selects a relevant extraction Actor, and inspects its input and output schema.
- Claude runs the Actor and uses the structured result in the next implementation step.
Outcome: The engineer gets real scraped data inside the coding session without building a one-off integration first.
Example 2: Docs plus Actor execution in one tool layer
- Team enables `actors` and `docs` tool groups.
- Claude searches Apify docs for setup guidance, then searches Actors for the exact workflow it needs.
- Claude combines documentation context with a live Actor run instead of relying on assumptions.
Outcome: Setup mistakes drop because the model can verify guidance against current docs and current actor metadata.
Example 3: Minimal production configuration
- Ops team restricts the hosted server to `actors,docs,apify/rag-web-browser`.
- Claude uses only the approved actor subset during production workflows.
- Team monitors Apify usage in console and expands the allowed tool list only when a new workflow clearly requires it.
Outcome: The MCP connection stays lean, predictable, and easier to govern over time.
Frequently Asked Questions
What is the Apify MCP Server?
The Apify MCP Server is Apify's official MCP integration that lets Claude Code and other MCP clients discover Actors, run them, access run results, and search Apify documentation through a single server.
How do I connect Apify MCP Server to Claude Code?
Apify documents two main options. The recommended hosted path uses `https://mcp.apify.com` with OAuth or a bearer token. If your client needs stdio, Apify documents a local setup using `npx -y @apify/actors-mcp-server` with `APIFY_TOKEN` in the environment.
What is the recommended Apify setup mode?
Apify explicitly recommends Streamable HTTP with OAuth for the hosted server because it is quick to set up and does not require exposing your API token directly. Local stdio is the fallback when the client does not support remote MCP servers well.
Can I limit which Apify tools are loaded?
Yes. Apify documents tool selection with query parameters such as `https://mcp.apify.com?tools=actors,docs,apify/rag-web-browser` or actor-specific selections like `https://mcp.apify.com?tools=apify/instagram-scraper,apify/google-search-scraper`.
What tools are enabled by default in Apify MCP Server?
Apify says the default setup includes essential tools for Actor discovery, documentation search, and the RAG Web Browser Actor. The docs also list default tools like `search-actors`, `fetch-actor-details`, `call-actor`, and `search-apify-docs`.
What are the main production best practices for Apify MCP?
Apify recommends explicitly selecting tools instead of relying on defaults, using the latest `@apify/actors-mcp-server` version for local stdio, monitoring API usage in Apify Console, and ensuring Node.js v18 or higher in local environments.
Missing a better tool match?
Send the exact workflow you are solving and we will prioritize a new comparison or rollout guide.